Scene Recognition through Visual Attention and Image Features: A comparison between SIFT and SURF Approaches